How Can Autistic Distress Be Missed by Services Designed to Detect It
Article two of three. Minds in Session.
This article discusses suicide and suicidal thinking in general terms. It contains no description of methods and no individual accounts. If you are struggling as you read, the sources of help listed at the end are there for you.
The first article in this series set out the evidence that suicidal thoughts and behaviours are markedly more common among autistic people than in the general population and that the contributors to that risk lie substantially in the environment and in service design rather than in autism itself.
This article addresses a further question. The findings have been replicated for a decade. Why does autistic distress continue to go unrecognised, including within services whose function is to recognise it?
Four factors operate and they compound one another.
The person has not been identified as autistic
In 2022, Sarah Cassidy and colleagues, with Simon Baron-Cohen as senior author, published a study in the British Journal of Psychiatry examining the coroners' inquest records of 372 people who died by suicide between 2014 and 2017 in two regions of England. In 10.8 % there was evidence in the record of a possible autistic profile or elevated autistic traits. Autism is estimated to affect between 1 and 2 % of the population. Only 0.5 % of the sample had a recorded autism diagnosis.
A smaller subgroup of 29 cases had additional interviews with bereaved relatives, and in twelve of those there was evidence of elevated autistic traits. That figure should be treated with caution given the size of the subgroup, the self-selection involved in who agreed to take part and the fact that those interviews were drawn from only one of the two regions; the 10.8 % from the full record review is the more dependable estimate. The authors recommended that coroners gather evidence on autism systematically at inquest, which at present they do not.
The clinical significance is straightforward. Once a clinician knows a person is autistic, features of the presentation that were previously difficult to interpret become intelligible and a range of adjustments and interventions becomes available that would not otherwise be considered. Without that information, the clinician is reasoning from an incomplete formulation and may reach a confident conclusion on the wrong basis.
Diagnostic delay is therefore a matter of patient safety rather than administrative convenience. It falls hardest on those whose presentation departs from the stereotype: autistic women and girls, academically able autistic people, those who have become practised at appearing to cope, and those from communities whose distress is already less readily believed.
Distress does not present in the expected form
Clinical judgement about risk relies heavily on a set of expected signals: reduced eye contact, flattened or tremulous voice, affect congruent with content, tears at the difficult question and a narrative delivered with pacing that marks it as painful.
Autistic people may not produce these signals, or may produce them in configurations that do not map onto the expected pattern. An account of a devastating experience may be delivered in an even, factual register because that is how the person organises and communicates information. Eye contact may be sustained and effortful precisely because it has been learned as an expectation. A question about hopelessness may receive a precise answer to the literal question asked, which is not the same as an answer to the question the clinician intended.
Camouflaging compounds this significantly. The 2018 study by Cassidy, Bradley, Shaw and Baron-Cohen established camouflaging as an independent predictor of suicidality after controlling for depression, anxiety and employment status. The behaviour that raises risk is also the behaviour that conceals it. Where a person's practised strategy is to appear to be managing, an assessment that relies on the appearance of managing will return a reassuring result.
Alexithymia adds a further constraint. On the meta-analytic estimate of Emma Kinnaird, Catherine Stewart and Kate Tchanturia at King's College London, approximately half of autistic people have marked difficulty identifying and describing their own emotional states. Such a person may be unable to give an accurate account of their internal condition irrespective of willingness. Asked to rate mood out of ten, they may have no reliable basis for the judgement. Asked whether things have deteriorated recently, they may not know.
Diagnostic overshadowing
Diagnostic overshadowing describes the attribution of new symptoms to an established diagnosis, so that they are absorbed rather than examined. Once autism is recorded, low mood becomes a feature of the autism; withdrawal becomes the person's baseline; self-injury is reclassified as stimming or as behaviour; exhaustion becomes longstanding temperament. A treatable mental illness, or a deteriorating and dangerous state, ceases to be investigated.
The 200 autistic adults surveyed by Camm-Crosbie and colleagues in 2019 reported this consistently, and the authors identified a lack of professional understanding of autistic people with co-occurring mental health difficulties as a dominant theme. Autistic people are simultaneously more likely to experience mental health difficulty and less likely to have it treated as such.
Overshadowing also operates in reverse. Autistic burnout, in the sense defined by Raymaker and colleagues, is regularly diagnosed and treated as depression. When the person does not improve, the record may show treatment resistance or poor engagement, when the accurate account is that an intervention was directed at the wrong target.
The assessment instruments were not designed for this population
Standard suicide risk assessment depends on a small number of questions, often administered under time pressure, and interpreted against norms derived from non-autistic populations. Several difficulties arise together.
The wording is often figurative. Items turn on formulations such as feeling like a burden, life not being worth living, or thoughts of ending it all. An autistic respondent may interpret these literally, find them too imprecise to answer accurately, or answer the words rather than the intended meaning.
The format presumes that a person can retrieve and report an internal state on demand, in an unfamiliar room, to a stranger, in the knowledge that the answer has consequences. For a substantial proportion of autistic respondents, that presumption does not hold or at least should not be assumed.
The risk factors themselves may differ. Camouflaging, sensory overload, autistic burnout, the loss of routine and the collapse of a valued structure carry weight in this population that standard instruments do not capture.
Cassidy and colleagues responded to this in 2021 with the Suicidal Behaviours Questionnaire for Autism Spectrum Conditions, published in Molecular Autism and developed with autistic people through cognitive interviewing before validation in autistic, possibly autistic and non-autistic community samples. The developers are explicit that the instrument was built for research and is not recommended for the assessment of future risk; and that scores are not directly comparable between autistic and non-autistic respondents. What it demonstrates is that standard item wording was not fit for this population and that better wording is achievable. Few services have yet acted on that finding.
The composite position
It is possible for an autistic person to be in serious difficulty, to seek help, to be assessed by several professionals, to answer every question honestly, and to be judged as low risk throughout. This does not require any individual failure of care. It follows from applying a system calibrated to one mode of expressing distress to that of a person who expresses it differently.
The remedies are correspondingly specific. Clinicians should hold the possibility of autism in mind, particularly where a presentation has not previously cohered. Questions should be plainly worded and concrete, with processing time allowed and understanding checked. Corroboration should be sought from those who know the person well rather than inferred from presentation alone. A composed account should be treated as information about the account. Finally, a diagnosis already held should not be permitted to absorb every subsequent difficulty.
The third article in this series sets out what the intervention evidence supports.
Sources of support
If you are struggling, you do not have to wait until things are worse to ask for help.
• Samaritans. Free, day or night, 365 days a year. Call 116 123, at any hour. The Welsh Language Line, 0808 164 0123, is open from 7 pm to 11 pm every day. A Relay UK service is available for d/Deaf and hard of hearing callers, and details are on the Samaritans website. Branches also offer face-to-face support; it is worth telephoning ahead to arrange a time.
• Shout. Free, confidential text support, 24 hours a day, which may suit you better if speaking on the telephone is difficult. Text SHOUT to 85258.
• PAPYRUS HOPELINE247, for anyone under 35 experiencing thoughts of suicide, and for anyone concerned about a young person. Call 0800 068 4141, text 88247, or email pat@papyrus-uk.org.
• Urgent NHS mental health support. In England and Wales, call 111 and select the mental health option. In Scotland, call NHS 24 on 111. In Northern Ireland, Lifeline is available on 0808 808 8000.
• Your GP, who can refer you into local services.
• In an emergency, call 999 or attend your nearest A&E department.
About the author
Esther John, Clinical and Forensic Psychologist, Minds in Session.
Minds in Session provides psychological assessment, therapy and organisational consultancy, with particular expertise in neurodiversity, leadership and workplace mental health.
References. Cassidy S, Au-Yeung S, Robertson A, et al. (2022) Autism and autistic traits in those who died by suicide in England. British Journal of Psychiatry 221(5):683-691. Cassidy S, Bradley L, Shaw R, Baron-Cohen S (2018) Risk markers for suicidality in autistic adults. Molecular Autism 9:42. Cassidy S et al. (2021) Development and validation of the suicidal behaviours questionnaire: autism spectrum conditions in a community sample of autistic, possibly autistic and non-autistic adults. Molecular Autism 12:46. Camm-Crosbie L, Bradley L, Shaw R, Baron-Cohen S, Cassidy S (2019) 'People like me don't get support'. Autism 23(6):1431-1441. Kinnaird E, Stewart C, Tchanturia K (2019) Investigating alexithymia in autism: a systematic review and meta-analysis. European Psychiatry 55:80-89. Raymaker DM et al. (2020) Defining autistic burnout. Autism in Adulthood 2(2):132-143.
